Wrote 3/25/2008 Rev. 7/5/2014 reposted
Without reliability of assessments you can’t have validity of the student’s gains or mastery of the subject. The reason for this is if the assessment tools are not reliable, as in you can't come up with the same results repeatedly. Then you can't find the level of proficiency of a student or a group of students. If this is true then you can’t use these tools with any validity to the ability of the student.
Test question must be statistically normalized before you can use them to determine if a student or group of students are meeting a level of proficiency. This is my concern, when we develop common end of quarter assessment to see if student are gaining knowledge needed to pass the Standardized test. I believe that these common assessments may not be a valid predictor of how the students would do on a standardised test. Also as we tested student and determined that some of the question were not reliable and then change the question by modifying it or removing them. we are changing are baseline and it would make it difficult to make any comparison between the groups of students being tested. Another issue of concern is that in the state of Michigan when they have done standardized testing, they are comparing school success by comparing different groups of students. If you compare different groups of student instead of the same cohort group. The data gain from this practice is of no practical use in determining educational best practices of a school and their effectiveness.